id	author	title	date	pages	extension	mime	words	sentences	flesch	summary	cache	txt
work_4ygq55srjzd6nallw4qqeo3tku	V. W. Feng	Patterns of local discourse coherence as a feature for authorship attribution	2013.0	8	.pdf	application/pdf	4912	560	69	We define a model of discourse coherence based on Barzilay and Lapata's entity grids as a stylometric feature for authorship attribution. test it against and in combination with standard features on nineteen booklength texts by nine nineteenth-century authors. makes when referring to an entity as to which grammatical role in the sentence the reference will appear role of an entity in the corresponding sentence: subject (S), object (O), neither (X), or nothing (–). entity grid and the corresponding coherence feature either the coherence feature set, the lexico-syntactic we hypothesized, our coherence features also significantly exceed the baseline in all cases, showing that Dickens versus Anne Brontë, the coherence features better than baseline, and the combined feature set Our experiments show that entity-based local coherence, by itself, is informative enough to be able local coherence is a stylometric feature. aggregated across all authors for each feature set Modeling local coherence: an entity-based approach.	./cache/work_4ygq55srjzd6nallw4qqeo3tku.pdf	./txt/work_4ygq55srjzd6nallw4qqeo3tku.txt